Semiautomatic assessment of endothelial density and morphology in organ-cultured corneas - potential predictors for transplantation suitability and clinical outcome?

Abstract

Background: The quality of the endothelial cell layer is a major criterion for the approval of organ-cultured human donor-corneas for transplantation. We wanted to compare the predictive capacities of initial endothelial density and endothelium cell morphology for the approval of donor corneas for transplantation and for the clinical outcome after transplantation.

Methods: The endothelial density and endothelium morphology in organ culture were examined by semiautomatic assessment of 1031 donor corneas. We performed a statistical analysis for correlations of donor-data and cultivation parameters regarding their predictive capacities for the final approval of donor corneas for transplantation and the clinical outcome of 202 transplanted patients.

Results: Corneal endothelium cell density proved to be the only parameter with a certain predictive capacity with regard to the final decision, whether donor corneas are suitable for transplantation - however, the correlation was low (area under the curve [AUC] = 0.655). Endothelial cell morphology lacked any predictive power (AUC = 0.597). The clinical outcome regarding visual acuity seemed to be largely independent from both corneal endothelial cell density and morphology. Sub-analyses on transplanted patients stratified for their diagnoses vindicated these findings.

Conclusions: Higher endothelial density (above a cut-off level of 2000 cells/mm²), as well as better endothelial morphology do not seem to be critical for transplant-corneal functionality in organ culture and up to 2 years after transplantation. Comparable long-term studies on graft survival are recommended to determine, whether the present endothelial density cut-off levels might be too stringent.

Keywords: Cell density; Clinical outcome; Corneal endothelium; Morphology.

Fig. 1

Anonymized printouts of exemplary semi-automated EAS-analyses on two corneas analyzed in our study. The left analysis shows a high quality endothelium, the right one an endothelium of poor quality. Automatically identified cells (marked by green or red dots within the target square of the microscopic photograph) before and after individual correction (numbers labeled by yellow and green frames respectively) and the cell density (based on corrected cell numbers [green dots] in cells/mm²; labelled in light blue frames) are shown at the upper left sides of the printouts, while morphology data (percentages of x-edged cells [green and red dots] after correction; labelled in dark blue frames) are shown at the upper right sides of the print outs

Fig. 2

Initial ECD/ECM and ECD/ECM–stabilities (= relative losses per day) in organ culture as predictors for transplantation release. ROC-Curves. Note that the areas under the curve values (AUCs; a value of 0.5 indicates the complete failure of the tested predictor, a value of 1 indicates a perfect prediction) are indicated within the graphs

Fig. 3

Initial ECD/ECM as predictors for the clinical outcome. ROC-curves. Above: Individual visual acuity (left graph) and objective refraction (measured light refraction by the cornea; right graph) 12 months after transplantation are displayed. The areas under the curve values (AUCs; a value of 0.5 indicates the complete failure of the tested predictor, a value of 1 indicates a perfect prediction) are indicated within the curves. Below: AUCs of the ROC-curves of all assessed tests are depicted as a function of time post transplantation (months)